Report 3

Jason Ola

2021-03-03

Import libraries

library(tidyverse)
library(readxl)
library(janitor)
library(lubridate)
library(ggrepel)
library(sf)

Exercises

Question 1

Let’s first load the data

geg_raw <- read_excel("data/TESEM0601573030089446.xlsx", 
    sheet = "Sheet 1", skip = 8)

Then we clean the data by removing unnecessary information and NA rows, we also pivot the table and change Germany name

geg_clean <- geg_raw %>% 
  slice(5:n()) %>% head(38) %>% 
  rename(country = time) %>% 
  select(-matches("\\.\\.\\.")) %>% 
  mutate_at(vars(-country),as.numeric) %>% 
  pivot_longer(cols = -country, 
               names_to = "year", 
               values_to = "gender_employment_gap") %>% 
  mutate(year = parse_date(year,"%Y")) %>% 
  drop_na() %>% 
  mutate(country = if_else(
    country == "Germany (until 1990 former territory of the FRG)",
    "Germany", country))

We split our data into 2 parts so we can plot each one

geg_mean_over_15 <- geg_clean %>% 
  group_by(country) %>% 
  mutate(mean_geg = mean(gender_employment_gap)) %>% 
  filter(mean_geg > 14.4)

geg_mean_under_15 <- geg_clean %>% 
  group_by(country) %>% 
  mutate(mean_geg = mean(gender_employment_gap)) %>% 
  filter(mean_geg <= 14.4)

Now we plot

geg_plot <- geg_mean_over_15 %>% 
  ggplot() +
  geom_line(aes(year,
                gender_employment_gap,
                color = country),
            show.legend = FALSE,
            size = 0.4)+
  geom_text_repel(data = geg_mean_over_15 %>% filter(year == max(year)),
                  aes(label = country,
                      x = year,
                      y = gender_employment_gap,
                      color = country),
                  size = 2.5,
                  show.legend = FALSE)+
  scale_color_manual(values = c("blue","red","brown","darkgreen",
                                "darksalmon","darkblue","darkred",
                                "magenta","black","pink","purple","orange"))+
  geom_line(data = geg_mean_under_15,
            aes(year,
                gender_employment_gap,
                fill = country),
            color = "gray",
            alpha = 0.3,
            size = 0.5)+
  labs(title = "Gender employment gap",
       subtitle = "Annual difference between the employment rates of mean",
       x = "Year",
       y = "Percentage of total population",
       caption = "Source : Eurostat")+
  theme_minimal()

geg_plot

Question 2

Let’s highlight a country, we choose Switzerland, we first filter what we need, then we add it on our plot

swiss_geg <- geg_clean %>% 
  filter(country == "Switzerland")

geg_clean %>% 
  filter(year >= as_date("2010-01-01")) %>% 
  ggplot() +
  geom_line(aes(year,
                gender_employment_gap,
                fill = country),
            size = 0.2,
            color = "grey")+
  geom_line(data = swiss_geg,
            aes(year,
                gender_employment_gap),
            size = 0.5,
            color = "blue")+
  labs(title = "Swiss gender employment gap",
       subtitle = "Annual difference between the employment rates of mean since 2010",
       x = "Year",
       y = "Percentage of total population",
       caption = "Source : Eurostat")+
  theme_minimal()

Question 3

Let’s now plot the data on a map, first we load our shape file

world <- st_read("data/naturalearth/ne_110m_admin_0_countries.shp")

## Reading layer `ne_110m_admin_0_countries' from data source `/Users/jasonola/Desktop/epfl_courses/viz/project-adscv-c1-s20-3268-2539/report3/data/naturalearth/ne_110m_admin_0_countries.shp' using driver `ESRI Shapefile'
## Simple feature collection with 177 features and 94 fields
## geometry type:  MULTIPOLYGON
## dimension:      XY
## bbox:           xmin: -180 ymin: -90 xmax: 180 ymax: 83.64513
## geographic CRS: WGS 84

Now we create the bbox limits

xlim <- c(-22,44)
ylim <- c(35,70)

Then we join the world shape with the data

world_data <- world %>% 
  left_join(geg_clean, c("SOVEREIGNT" = "country"))

Finally we plot both

ggplot()+
  geom_sf(data = world_data,
          mapping = aes(fill = gender_employment_gap),
          size = 0.1)+
  coord_sf(xlim = xlim,
           ylim = ylim)+
  scale_fill_continuous(na.value = "lightgray")+
  theme_void()+
  labs(title = "Gender employment gap in EU",
       subtitle = "Annual difference between the employment rate of men and women aged 20 - 64",
       caption = "Source : Eurostat")+
  theme(legend.title = element_blank())

Question 4

Let’s now make a quantile column to use in our map, first we create the quantiles

quantiles <- world_data %>% 
  pull(gender_employment_gap) %>% 
  quantile(probs = c(0,0.25,0.5,0.75,1),
           na.rm = TRUE)

Then from these quantiles we create the labels

labels <- tibble(
  lab1 = quantiles,
  lab2 = lead(lab1)
) %>% 
  slice(1:n()-1) %>% 
  mutate(lab2 = paste0(lab2, "%")) %>% 
  mutate(label = paste(lab1,lab2, sep = " to "))

We now put these in our data

world_data <- world_data %>% 
  mutate(percentages = cut(gender_employment_gap,
                           breaks = quantiles,
                           labels = labels$label))

And we plot

ggplot()+
  geom_sf(data = world_data,
          mapping = aes(fill = percentages),
          size = 0.1)+
  coord_sf(xlim = xlim,
           ylim = ylim)+
  viridis::scale_fill_viridis(discrete = TRUE,
                              option = 8,
                              direction = -1,
                              na.value = "lightgray",
                              begin = 0.2,
                              end = 0.8)+
  theme_void()+
  labs(title = "Gender employment gap in EU",
       subtitle = "Annual difference between the employment rate of men and women aged 20 - 64",
       caption = "Source : Eurostat")+
  theme(legend.title = element_blank())

Question 5

Let’s now annotate our map

world_data <- world_data %>% 
  group_by(NAME) %>% 
  mutate(mean_geg = round(mean(gender_employment_gap),0)) %>% 
  ungroup()

world_data %>% 
  ggplot()+
    geom_sf(aes(fill = percentages),
            size = 0.1)+
    coord_sf(xlim = xlim,
             ylim = ylim)+
    viridis::scale_fill_viridis(discrete = TRUE,
                                option = 8,
                                direction = -1,
                                na.value = "lightgray",
                                begin = 0.2,
                                end = 0.8)+
    geom_sf_text(aes(label = mean_geg),
                 color = "gold",
                 size = 3,
                 na.rm = TRUE)+
    theme_void()+
    labs(title = "Gender employment gap in EU",
         subtitle = "Annual difference between the employment rate of men and women aged 20 - 64",
         caption = "Source : Eurostat")+
    theme(legend.title = element_blank())

Question 6

Let’s now separate in 2 different times with facets, we filter our desired years first

world_data %>% 
  filter(year %in% c(ymd("2014-01-01"),ymd("2018-01-01"))) %>% 
  mutate(year = year(year)) %>% 
  ggplot()+
  geom_sf(aes(fill = percentages),
          size = 0.1)+
  coord_sf(xlim = xlim,
           ylim = ylim)+
  viridis::scale_fill_viridis(discrete = TRUE,
                              option = 8,
                              direction = -1,
                              na.value = "lightgray",
                              begin = 0.2,
                              end = 0.8)+
  facet_wrap(vars(year))+
  theme_void()+
  labs(title = " Gender employment gap in EU",
       subtitle = " Difference between 2014 and 2018 \n",
       caption = "Source : Eurostat")+
  theme(legend.title = element_blank())

Question 7

Let’s now highlight the difference between 2014 and 2018, we first need to get the difference between 2 different years in another table because we have to do some pivot wider.

diff_world <- world_data %>% 
  filter(year %in% c(ymd("2014-01-01"),ymd("2018-01-01"))) %>% 
  mutate(year = year(year)) %>% 
  select(year,gender_employment_gap,NAME) %>% 
  as_tibble() %>% 
  pivot_wider(names_from = year, values_from = gender_employment_gap, names_prefix = "y_") %>% 
  mutate(diff = y_2018 - y_2014) %>% 
  select(NAME,diff)

Then we join our new column to the world data

world_data <- world_data %>% 
  left_join(diff_world, by = "NAME")

We again do some quantile operation on the diff column

diff_quantiles <- world_data %>%
  pull(diff) %>% 
  quantile(probs = c(0,0.25,0.5,0.75,1),
           na.rm = TRUE)

diff_labels <- tibble(
  lab1 = diff_quantiles,
  lab2 = lead(lab1)
) %>% 
  slice(1:n()-1) %>% 
  mutate(lab2 = paste0(round(lab2,0), "%")) %>% 
  mutate(label = paste(round(lab1,0),lab2, sep = " to "))

world_data <- world_data %>% 
  mutate(diff_percentages = cut(diff,
                           breaks = diff_quantiles,
                           labels = diff_labels$label))

Finally we plot the results with a diverging palette

ggplot()+
  geom_sf(data = world_data,
          mapping = aes(fill = diff_percentages),
          size = 0.1)+
  coord_sf(xlim = xlim,
           ylim = ylim)+
  scale_fill_brewer(palette = "PuOr", direction = -1)+
  theme_void()+
  labs(title = "Gender employment gap in EU",
       subtitle = "Difference between 2014 and 2018",
       caption = "Source : Eurostat")+
  theme(legend.title = element_blank())

Question 8

Let’s change projections now

esri_world_data <- st_transform(world_data, crs = 'ESRI:54030')

We have to change our border box in order to plot our Europe area

xlim_rob <- c(-1733617, 3443854)
ylim_rob <- c(3765900,7518072)

Now we can plot our map

ggplot()+
  geom_sf(data = esri_world_data,
          mapping = aes(fill = diff_percentages),
          size = 0.1)+
  coord_sf(xlim = xlim_rob,
           ylim = ylim_rob)+
  scale_fill_brewer(palette = "PuOr", direction = -1)+
  theme_void()+
  labs(title = "Gender employment gap in EU",
       subtitle = "Difference between 2014 and 2018",
       caption = "Source : Eurostat")+
  theme(legend.title = element_blank())

Question 9

top5_hi <- world_data %>% 
  filter(year == "2018-01-01") %>% 
  slice_max(order_by = gender_employment_gap, n = 5) %>% 
  pull(NAME) %>% 
  glue::glue_collapse(", ", last = " and ")

top5_low <- world_data %>% 
  filter(year == "2018-01-01") %>% 
  slice_min(order_by = gender_employment_gap, n = 5) %>% 
  pull(NAME) %>% 
  glue::glue_collapse(", ", last = " and ")

top5_low_diffs <- world_data %>% 
  filter(year == "2018-01-01") %>% 
  slice_min(order_by = diff, n = 5) %>% 
  pull(NAME) %>%
  glue::glue_collapse(", ", last = " and ")

The top 5 countries with the highest gender employment gap are : Turkey, Greece, Italy, Romania and Hungary
The top 5 countries with the lowest gender employment gap are : Lithuania, Finland, Latvia, Sweden and Norway
The top 5 countries which the employment gap decreased the most in percentage points are : Luxembourg, Turkey, Czechia, Falkland Is. and United Kingdom